- SaaS
- AI Business Automation
- ML & Data Science
- B2B Marketing
Users leaving before they find the good part
A product with strong retention past ninety days loses a large share of signups in the first week, to a generic drip sequence that sends the same day three email whether or not the user got anywhere. Triggering on behaviour instead of on time fixes the mismatch.
Representative engagement. This describes a pattern we build rather than one named client: the situation that produces it, how we approach it, and the range of outcomes that kind of work lands in. Figures are stated as ranges or targets, never as a measured result for a specific customer. Our named client work is on the work index.
- Industry
- SaaS
- help at the point of the stall, not on a schedule
- Behaviour triggered
- 90 day retention lift (target range for this pattern)
- 10 to 25 pts
The problem
Time based onboarding assumes every user progresses at the same rate, which none of them do. The user who finished setup gets encouraged to finish setup, and the one who stalled on step two gets a message about step five. The deeper problem is that nobody has defined activation. Without a specific action that predicts retention, the sequence is aimed at nothing in particular and cannot be evaluated, so it never improves. Adding more emails is the usual response, and it makes things worse, because the users who are doing fine are the ones most likely to unsubscribe from them.
What we built
Activation gets defined first, from the data: the specific action, or short sequence of actions, that separates users who are still there at ninety days from those who are not. Everything else is aimed at that. Messaging then triggers on where a user actually is. Someone who stalls at a step gets help with that step, in the product as well as by email. Someone progressing well is left alone, which is an underrated intervention. In app guidance carries the load that email cannot, since the user is already in the place where the action happens. Everything is instrumented, so each trigger can be evaluated against activation rather than against open rate.
What changed
Time to first value shortens because help arrives at the point of the stall rather than on a schedule, and the users who do not need help stop being interrupted. The retention figure below is the range this pattern is scoped against rather than a measured result. What the work reliably produces is the ability to tell which intervention did anything, which most onboarding programmes cannot answer at all.
Built with
- TypeScript
- PostHog
- PostgreSQL
- Next.js
- Resend